|sleep apnoea project|
My main task for the diploma project was the investigation of computational techniques
for decision support in the diagnosis of sleep apnoea. Sleep apnoea is a disease
with obvious symptoms such as daytime sleepiness and diffuculties to concentrate.
The reason for these symptoms are apnoeas - stop of breathing - because of obstruction
of the pharynx and larynx during the night. The human body reacts on the resulting
low blood oxygen saturation with a "reconfiguration" of the participating muscles
which leads to a disturbance (arousal) of sleep. |
The current diagnosis of sleep apnoea includes recording of a set of physiological signals during the night of patients in a sleep laboratory. The recordings have to be marked manually which is a time-consuming process. The most important information for the diagnosis lies in recorded airflow, and in characteristics of the subject's sleep, i.e. the sequence and percentage of different sleep stages. I was interested in amplitude features of the electro-physiological signals (electro-encephalography, electro-oculography, electro-myography) which are used for sleep staging. I tried to find answers for two questions:
1) Is the creation of distinct sleep stage classes possible by using amplitude features?
2) Do amplitude features differ between normal and pathological subjects?
The applications for signal and amplitude feature viewing, and the toolboxes for
signal loading, signal transformation, amplitude feature extraction, statistics creation,
and fuzzy membership set creation, fuzzy inference system, and some additional
tools were all implemented using Matlab 5.3 (The Mathworks Inc.). The code is
free software and is subject to the GNU public license. Please cite my diploma thesis
if software is used in academical environment: |
Andreas Gerst (2005). Computational Techniques for Decision Support in the Diagnosis of Sleep Apnoea. Diploma thesis. Medical Faculty, University of Heidelberg.
The results of the project can be read in my diploma thesis which was submit
end of january and marked in may.
|manuals and additional files|
|© 2006 by agerst||home|